agentic.markets

Ground-truth Indian financial data
for AI agents.

A hosted MCP server for India + EM finance. NSE, BSE, AMFI, MCA, SEBI, Yahoo, IBBI — one endpoint, 45 tools, full provenance on every response.

45tools

12.5M+rows

5,506NSE-listed equities

14,351mutual fund schemes

486Kcorporate filings

2.4yprice history

Connect in 30 seconds

In Claude Desktop, open Settings → Connectors → Add custom connector. Use:

Name: AgenticMarkets
URL:  https://mcp.agenticmarkets.dev/mcp

Leave the OAuth fields blank — the server publishes a discovery document and runs Dynamic Client Registration automatically. Click Approve when the consent page opens.

claude mcp add --transport http --scope user agenticmarkets \
  https://mcp.agenticmarkets.dev/mcp \
  --header "Authorization: Bearer am_live_..."

Get an API key by emailing hello@agenticmarkets.dev. Trial keys are unmetered for the first month.

pip install agenticmarkets

from agenticmarkets import AgenticMarketsClient

am = AgenticMarketsClient(api_key="am_live_...")

quote = am.get_quote(ticker="RELIANCE")
print(f"{quote.entity.primary_symbol}: ₹{quote.close} on {quote.trade_date}")
print(f"source: {quote.meta.sources[0].name}, run: {quote.meta.sources[0].ingest_run_id}")

# 1. initialize a session
INIT=$(curl -fsS -X POST https://mcp.agenticmarkets.dev/mcp \
  -H "content-type: application/json" \
  -H "accept: application/json, text/event-stream" \
  -H "Authorization: Bearer am_live_..." \
  -d '{"jsonrpc":"2.0","id":1,"method":"initialize",
       "params":{"protocolVersion":"2025-06-18","capabilities":{},
                 "clientInfo":{"name":"x","version":"0"}}}' -i)
SID=$(echo "$INIT" | grep -i 'mcp-session-id:' | awk '{print $2}' | tr -d '\r')

# 2. call any tool
curl -X POST https://mcp.agenticmarkets.dev/mcp \
  -H "Authorization: Bearer am_live_..." \
  -H "mcp-session-id: $SID" \
  -d '{"jsonrpc":"2.0","id":2,"method":"tools/call",
       "params":{"name":"getQuote","arguments":{"ticker":"RELIANCE"}}}'

Why this exists

Indian and emerging-markets finance data is fragmented. NSE, BSE, AMFI, MCA, SEBI, IBBI, regulators, exchanges, vendors — an analyst chasing one ticker traverses a dozen sources. Bloomberg/Refinitiv don't ship to most teams that need it; Polygon and friends don't cover India.

AgenticMarkets is the layer in between. One MCP endpoint, every Indian listed equity and mutual fund, every corporate filing back to 2021, F&O chains, shareholding patterns, insider trades, credit ratings, earnings call transcripts, ESG signals, sector indicators. Each response carries the canonical entity, the source, the ingestion run id, and a SHA-256 row hash — so an AI agent's analysis is always traceable.

Tools

Equities — NSE

getQuotelatest daily OHLC, single ticker

getOhlcdate-range bars, single ticker

getYahooQuote5-year alt feed, NSE+BSE

getMarketCapdaily mcap snapshots

getDeliverydelivery percentage

getHiLoEvents52w breakouts/breakdowns

Equities — BSE

getBseQuotelatest BSE daily quote

getBseOhlcdate-range BSE bars

Derivatives

getFnoContractsNSE F&O bhavcopy

getBseFnoBSE F&O bhavcopy

Mutual funds & ETFs

getNavAMFI NAVs

getEtfETF reference data

Indices

getIndexNIFTY/SENSEX history

getIndexConstituentsindex members

getMarketStatsgainers/losers/breadth

Filings & disclosures

getFilingNSE corporate announcements

getBseFilingBSE corporate announcements

searchFilingsfull-text across filings

getFilingFullTextcomplete OCR'd text

getAnnualReportannual report PDFs

getFinancialResultsquarterly results

getNseCircularregulatory circulars

Fundamentals & research

getFundamentals5-year annual statements

getFundamentalsSnapshotTTM rolling

getRatiosPE/PB/ROE/ROCE etc.

getPriceStats52w hi/lo, MA, beta

getConcallearnings call transcripts

getPeerscomparable tickers

Governance & ownership

getCompanyentity master record

searchCompaniesfull-text on entities

getQuoteMetadataF&O eligibility, lot size

getShareholdingquarterly pattern

getInsiderTradesSEBI PIT disclosures

getBulkDealsblock trades

getPromoterPledgepledge disclosures

getBoardMeetingmeeting calendar

getCreditRatingCRISIL/ICRA/CARE

getAuditorstatutory auditor records

getCirpIBBI bankruptcy filings

getCorporateActionsdividends, splits, bonuses

News & macro

getNewsfinancial news articles

getSectorSignalFADA, Vahan, FX, etc.

getMacroIndicatorGDP, CPI, repo

Composite & meta

getCompanyContextone-call bundle composite

server_infoversion + tool catalog

→ full tool reference with input/output schemas

Provenance is the contract

Every response carries:

data — the answer, validated against a Zod schema published in our SDK.
meta.sources[] — the source name (e.g. nse_cm_bhavcopy, sebi_pit, amfi), the source URL, the ingestion timestamp, and the ingestion run UUID.
meta.row_hash — SHA-256 over the canonical-JSON of data. Two clients querying the same row can verify they got the same bytes.
meta.freshness_sla_seconds — how stale this datum is allowed to be before we'd consider it breached. Compare to meta.served_at - meta.sources[*].ingested_at.

This is the metadata an AI agent needs to make verifiable decisions. Trust, but verify.

Rate limits & auth